APIC Webinar: Ethics and Strategy: Managing AI in Healthcare for a Resilient, Innovative Future

Good morning, everyone. My name is Kyle Atkins, and I’m honored to speak with you today about Ethics and Strategy: Managing AI in Healthcare for a Resilient, Innovative Future. We’ll explore how infection preventionists can apply artificial intelligence while keeping ethics and resilience at the forefront. Let’s dive in together.

This talk covers four essential areas: First, how to develop a thoughtful AI governance strategy. Second, how to minimize bias and risk in AI tools. Third, how to balance innovation with trust, so new technology builds confidence rather than erodes it. Finally, we’ll walk through real-world case studies to show how these principles play out in practice. By the end, you’ll have concrete steps to take back to your organization.

We’re living through an extraordinary adoption wave. A 2024 survey found that 71 % of U.S. hospitals are already using predictive AI tools, up from 66 % the previous year . Hospitals are not only adopting AI—most are actively evaluating their models for accuracy, bias, and ongoing performance . This isn’t about whether we will use AI; it’s about how we can do so responsibly and ethically, especially in infection prevention.

Let’s define AI in simple terms. Artificial intelligence is a system designed to learn patterns from data and make predictions. It’s not magic, and it’s not sentient. AI is a tool we train with data to support our judgment. The goal is to augment our clinical intelligence—not replace it. Keeping that distinction clear helps set realistic expectations for what AI can and can’t do.

As we consider AI, we need to acknowledge both sides. On the promise side, AI can increase efficiency, generate deeper insights, and speed up workflows dramatically. On the peril side, AI can amplify bias, compromise privacy, or be misused if not carefully governed. Today’s talk is about how to maximize those benefits while keeping the risks in check.

A 2023 study asked whether AI chatbots could help draft patient responses. Reviewers compared answers written by physicians with answers generated by ChatGPT and found that 78.6 % preferred the chatbot responses for quality and empathy . Chatbot answers were longer, higher quality, and more empathetic . This shows that AI can sometimes outperform humans in narrow tasks like drafting responses, but it raises a critical question: How do we preserve oversight and ethical standards when AI starts to outshine us?

Here’s a reminder of our guiding principle: AI may outshine us in empathy and efficiency, but without ethics, it cannot outlast us in trust. No matter how advanced AI becomes, trust is earned through transparent, fair, and accountable practices.

Our approach to AI should always start with ethics. Four pillars guide us: Transparency, so people understand how AI reaches conclusions; Accountability, because humans still make the final decisions; Bias Mitigation, through continuous testing and auditing ; and Patient Autonomy and Trust, ensuring AI supports informed consent rather than replacing it . Ethics aren’t a barrier to innovation; they’re the foundation that enables innovation.

When AI goes wrong, the consequences can be serious. A population health algorithm once used healthcare costs as a proxy for patient need. It inadvertently referred far fewer high-risk Black patients for extra care. The model’s cost-based proxy hid racial bias, there was no governance, and nobody audited the outputs. The result? Undermined trust, accountability gaps, and inequitable care. This case reminds us that even well-intentioned models can fail ethically without proper oversight.

Contrast that with Viz.ai’s stroke triage platform for large-vessel occlusions. Here, AI flags potential strokes and notifies teams in real time. At one center, time from hospital arrival to groin puncture decreased by about 39 % off-hours, accelerating life-saving treatment. Clinicians remained the decision-makers, and the metrics were transparent. This success story shows that when AI is carefully implemented—monitored for bias, used as a communication tool, and supported by governance—it can save lives while maintaining trust.

Which brings us to AI governance. Governance is the bridge between ethical intent and practical outcomes. It’s the system that ensures AI aligns with clinical values, regulations, and patient safety—so we know who is responsible, what metrics are monitored, and how we correct course when things go wrong.

I like to think of ethics as the ‘why’ and governance as the ‘how.’ Ethics gives us the reason; governance gives us the process. It connects the values we believe in—like fairness and transparency—to the outcomes we need—like accurate predictions and patient safety. AI governance makes sure ethics aren’t just abstract ideals but guiding principles in day‑to‑day work.

Effective governance requires the right people. Oversight committees should include clinical leaders, supply chain/value analysis experts, IT, compliance, and leadership. Defined roles ensure someone is responsible for approving AI tools, monitoring performance, and reporting back. Leadership accountability via an executive sponsor gives the governance team the authority it needs. AI governance is a team sport.

How do we put governance into practice? Create committees that review AI initiatives and ensure compliance. Monitor bias and privacy through routine audits and subject-matter experts. And align with national guidelines, like the HHS AI Playbook, which outline principles for trustworthy AI use. Governance isn’t about adding bureaucracy; it’s about standardizing processes so innovation can thrive safely.

Several tools make governance tangible. An AI Implementation Checklist guides adoption and scaling, ensuring you define objectives, identify responsible teams, and set guardrails. A Risk Assessment Matrix scores each project on potential harm and benefit, so you can prioritize oversight. Position statements and toolkits from professional groups such as AHVAP or the HHS provide templates and best practices to ensure consistency and compliance.

Imagine a Venn diagram of Innovation, Trust, and Accountability. True resilience lies at their intersection. We can encourage innovation through pilot programs and safe testing environments. We build trust with transparent communication—letting patients, clinicians, and administrators know what AI is doing. And we maintain accountability by celebrating wins and acknowledging when things fall short. When all three converge, we have a sustainable, ethical AI program

Let’s look at AI applied to infection surveillance. A 2024 study tested ChatGPT‑4 and a local Mixtral model on six National Healthcare Safety Network scenarios for CLABSI and CAUTI . Both models correctly classified all six cases when the data were complete . Errors only happened when key information—like device insertion dates—was missing . The lesson? AI can accurately augment surveillance, but data quality and human oversight remain crucial.

Another example comes from antimicrobial stewardship. In 2025, a hospital implemented an AI tool that predicted patients’ risk of C. difficile infection. Among high-risk patients, broad-spectrum antibiotic use decreased by 16.8 % . However, overall C. diff rates didn’t change . This shows AI can sharpen our interventions, but it’s not a silver bullet; clinical engagement and workflow integration are essential for impact.

Beyond clinical care, AI can streamline operations. Predictive analytics for inventory management have reduced supply waste, cut expired items by about 20 %, and projected annual savings of around $2 million in modeled hospital systems. By forecasting demand and preventing stock-outs, AI ensures critical infection-prevention materials are on hand without waste. This frees up resources for patient care.

AI works best when it’s not siloed. Integrating infection prevention, pharmacy, IT, and operations means shared data pipelines and unified governance. Cross‑departmental integration improves communication, reduces workflow interruptions, and speeds up responses to alerts. Hospitals that adopt enterprise-wide AI strategies report higher staff trust and accuracy across teams. In other words, integration multiplies AI’s value.

Now I’d love to hear from you. Please use your phones to scan the QR code or visit the poll link and answer: ‘How are you currently using AI in your infection prevention or clinical work?’ Are you exploring, experimenting, integrating, or scaling AI? Your responses will help us understand where the community stands and tailor our discussion accordingly

When you go back to your institution, ask these questions: How can AI support but not replace your infection prevention expertise? Which data sources are safe and appropriate for training AI? Where could AI meaningfully enhance manual reporting or surveillance? How will you measure success? And who will be accountable for governance and oversight? These questions help convert abstract ideas into actionable plans.

Here are five concrete steps you can take next week:

1. Form an AI governance working group—include IT, value analysis, compliance, and clinical leaders.

2. Review your current AI tools and partners for bias, transparency, and accountability.

3. Introduce an AI implementation checklist to standardize adoption.

4. Educate your team with a 30‑minute AI ethics primer.

5. Align your programs with national frameworks (like the HHS AI guidelines or AHVAP position statements).

   By tackling these tasks, you’ll move AI from concept to practice.

I want to leave you with this: ‘Artificial intelligence will not save healthcare. Leaders will. Ethics plus strategy equals resilience.’ AI is a powerful tool, but without ethical leadership and strategic vision, it can do more harm than good. The future depends on how we lead

Thank you for joining me on this journey. If you’d like to continue the conversation, here’s my contact information:

• Email: [email protected]

• LinkedIn: linkedin.com/in/ka7735

• Website: veritastech.io

Feel free to scan the QR code for additional resources and references. I appreciate your time and look forward to the innovative and ethical AI work you’ll do in your organizations.